We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.ST

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Statistics Theory

Title: Convergence Rates of Variational Posterior Distributions

Abstract: We study convergence rates of variational posterior distributions for nonparametric and high-dimensional inference. We formulate general conditions on prior, likelihood, and variational class that characterize the convergence rates. Under similar "prior mass and testing" conditions considered in the literature, the rate is found to be the sum of two terms. The first term stands for the convergence rate of the true posterior distribution, and the second term is contributed by the variational approximation error. For a class of priors that admit the structure of a mixture of product measures, we propose a novel prior mass condition, under which the variational approximation error of the mean-field class is dominated by convergence rate of the true posterior. We demonstrate the applicability of our general results for various models, prior distributions and variational classes by deriving convergence rates of the corresponding variational posteriors.
Subjects: Statistics Theory (math.ST); Machine Learning (stat.ML)
Cite as: arXiv:1712.02519 [math.ST]
  (or arXiv:1712.02519v4 [math.ST] for this version)

Submission history

From: Chao Gao [view email]
[v1] Thu, 7 Dec 2017 07:30:16 GMT (55kb,D)
[v2] Thu, 14 Dec 2017 02:15:27 GMT (55kb,D)
[v3] Fri, 2 Feb 2018 17:38:55 GMT (57kb,D)
[v4] Mon, 17 Jun 2019 04:44:38 GMT (134kb,D)

Link back to: arXiv, form interface, contact.